DM: RE: Prunning a tree

[Tom Dinsmore]

Are you really trying to build a tree from fourteen cases, or is the 
data
shown in your posting simply a sample from the dataset?

What pruning rule are you using?

--------------------------------------------------------------
Thomas W. Dinsmore tdinsmore@exapps.com
Engagement Manager
Exchange Applications
One Lincoln Plaza
89 South Street
Boston MA 02111
voice:   617-737-2244 x556
fax:      617-443-9143
mobile: 617-512-7297



-----Original Message-----
From: fardin akbaryan [mailto:fakbarya@ucalgary.ca]
Sent: Monday, January 25, 1999 3:59 PM
To: datamine-l@nautilus-sys.com
Subject: DM: Prunning a tree



    In  my research I am using ITI method proposed bu Utgoff. To test
how the method works, I used the original source code available 
through
internet. The data applied are as follow:

sunny,75,70,true,+.
sunny,80,90,true,-.
sunny,85,85,false,-.
sunny,72,95,false,-.
sunny,69,70,false,+.
overcast,72,90,true,+.
overcast,83,78,false,+.
overcast,64,65,true,+.
overcast,81,75,false,+.
rain,71,80,true,-.
rain,65,70,true,-.
rain,75,80,false,+.
rain,68,80,false,+.
rain,70,96,false,+.
In this set of data, +/- represent the classes and the two attributes
are continuouse and the rest are discrete. There is no missing
values.The tree constructed by the algorithm has 13 nodes. When the 
tree
is pruned by the ITI, all of nodes up to the root are removed. In
another word, I'll get a single node for the whole data set which
doesn't look fine. Would somebody please give me an idea about this 
type
of output? As far as the correctness of my program is concerned, I use
the Utgoff's original source and nothing more.

    Regards

--


 Fardin Akbaryan
 Chemical and Petroleum Eng. Department
 University of Calgary
 2500 University Drive NW
 Calgary T2N 1N4, Alberta, Canada
 Tel: (403) 220-7409
 Fax: (403) 284-4852